Hand-rolled numerics · seeded & reproducible
Model the loss
Frequency and severity distributions, frequency-severity and pre-sampled loss models, market models and a seeded Monte-Carlo sampler — every draw hand-rolled from core types (inverse-CDF, Knuth / transformed-rejection Poisson, Marsaglia-Tsang Gamma), no MathNet. Depends on the Reinsurance domain.
What ships
- Distribution family — frequency (Poisson, NegativeBinomial, Binomial, the unified Panjer, DiscreteUniform, Logarithmic) and severity (Pareto, ParetoLimited, TruncatedPareto, LogNormal, MixedExponential, Exponential, Gamma), each a value record with a hand-rolled seeded
Sample(Random)and analyticMean/Variance. - Loss models — the frequency-severity compound model and a pre-sampled model over time-series data cubes.
- Market models and a seeded Monte-Carlo sampler with a discrete-distribution estimator.
The reinsurance suite
This plugin is the stochastic engine of a three-part suite: Reinsurance defines the contracts whose sections these losses feed, Pricing binds the two and simulates the program — and the whole data model is browsable type by type in the Data Modelling explorer.